The meat computers inside our skulls have a network, when working together on a team. The information stored on this network can be called institutional knowledge.
The way the work gets done.
When we all come together to do a project, we learn things about the work. That institutional knowledge is what makes our future collaborations smoother, faster, and higher quality.
If you fire your human staff, and replace them with agents, you may find that the only people who could train those agents on how to do the work are the humans with institutional knowledge.
That’s what Klarna found out – they cut their staff of 3000 down to 2300, and tried to replace them all with AI. They ended up having to hire everybody back.
I was Queenstown and Wanaka earlier this month, delivering AI Training workshops to corporate teams. This is a non-AI-edited photo. (I like to share those once ina while, too, to remind you I’m still human.)
During the Ai202: AI for Knowledgebases workshop, we discussed the challenges of digitising this institutional knowledge. Part of the challenge is architectural: creating the right permissions structure, so the right people have access to the right levels of information.
The second challenge is ingestion – making sure new data continually contributes. To do this, you need ongoing human input to adjudicate what is and isn’t the right way to do things.
This is the work of the future: training AI agents to make human choices.
To train AI agents well, to continually improve over time, you need an ongoing, growing database of those choices.
That’s the Company Brain.

🗄️ “Companies need to turn their workflows, domain knowledge, and accumulated judgment into AI systems that improve with each use. The companies that build this early will have an advantage that is hard to replicate, regardless of any new individual model capability.”
- Satya Nadella, CEO of Microsoft
💡 A Company Brain is connected context in action.
Your ability to capture digital residue of institutional knowledge as it is created, and query it as needed, will determine your organisation’s thinking capacity.
Recording your meetings and automating the transcripts is only the first step. What does your Company Brain do with all that information?
- How is it ingested, processed, and refined?
- When wisdom is extracted from a transcript, where does it go?
- How does it inform choices in the future?
That unspoken structure that currently lives in the network of meat computers in the skulls of your team – that’s what Zoe Scaman calls Heartwood.
🤖 Building a Company Brain
There are a variety of tools that can be used to make and maintain a Company Brain:
- Obsidian
- Airtable
- Notion
- Google Drive
- LLM-Wiki
- GBrain
At its core, you want a series of markdown files that can reference one another quickly.
Who manages the Company Brain? In an AI Working Group, that person’s role is The Skull. They protect the Company Brain, help it to grow, and make sure the right people have access to it.
You could use a RAG, Retrieval Augmented Generation, and you could code it all up yourself. Or you can go with a tool like Caitlyn.AI, and have them set it up for you.
If you want a specific recommendation for your business, you could schedule an AI Agent Assessment with me.
📈 Relevant Statistics
- McKinsey Global Institute found in their study of knowledge workers, The Social Economy, that 19% of time is spent finding information.
- Asana’s survey of 9615 global workers found that 60% of time is spent on work about work.
- Gartner estimates that poor data quality costs $12.9million/year.
- In Redis’ recent report, the State of Context Engineering, 94% of organizations said that compounding system intelligence is essential for production-grade maturity. However, only 4% have reached the stage where that actually happens.
🧠 📦 Game: Brain in a Box
GOAL: Connect a pile of documents into a searchable Company Brain.
RULES: Download a document collection, upload into a Project (for Claude/ChatGPT) or a Notebook (for Gemini/Copilot), and ask questions.
FEEDBACK: How can this evolve beyond a search box, into interconnection?
Turn up your reasoning, upload a document dump, and enter the following prompts sequentially:
- Tell me what your project context contains: key people, projects, decisions, dates, topics, and recurring themes. Cite the source for each.
- Find 5 important connections across multiple documents that would be difficult to notice by reading them individually. Show me the evidence for each connection.
- Turn these documents into a Company Brain. Summarize what this organization knows, what decisions it has made and why, lessons it has learned, unresolved questions, and 10 valuable questions I can now ask this brain.
If you want some advice on structuring your own Company Brain, schedule an AI Agent Assessment. I’ll interview you in our first 45-minute call, then write you a report, and we’ll go over it in our second call.
🛠️ Expert Interview: Michael Philpott
[14:01]
👓 What I’m Reading
- Heartwood by Zoe Scaman – “Heartwood is the dense inner wood of a tree. It forms slowly, over decades. You can’t see it from outside. And once it’s cut out, you can’t grow it back on any useful timescale.In an institution, heartwood is the accumulated judgment, relational memory, ethical stance and tacit knowledge that distinguishes one organisation from another. It’s what the organisation is made of – not what it does.”
- Inside OpenAI’s Reboot by Alex Heath for Time Magazine – “The upswing is more fun after the downswing.”
- The Castlereagh Statement – A cross-sector call to action on Australian education and training in the age of AI
🗓️ Upcoming Events
Premium webinar for members of the AI Campus:
![]() |
Friday, September 25 at 11:00 AM GMT+12
Ai206 – AI for Creativity Generate New Ideas Use AI as a creative collaborator. This session helps you break through creative blocks, explore new directions, and accelerate ideation across writing, design, and problem-solving. Generate and expand ideas with AI co-creation tools. Explore alternative perspectives and creative directions. |


Leave A Comment